Can you suggest ways we could investigate slow rtmp streams? We stream them to another machine in the same cluster, and send a POST just before, but the stream often arrives 4-10 seconds after the POST.

At minimal load conditions